[kvm-unit-tests PATCH v6 00/14] arm64: EL2 support

[Andrew Jones]

On Fri, Jan 23, 2026 at 07:30:15PM +0000, Joey Gouly wrote:
> On Fri, Jan 23, 2026 at 01:23:41PM -0600, Andrew Jones wrote:
> > On Fri, Jan 23, 2026 at 04:50:39PM +0000, Joey Gouly wrote:
> > > Hi all,
> > > 
> > > This series is for adding support to running the kvm-unit-tests at EL2.
> > > 
> > > Changes since v5[1]:
> > > 	- make debug bp/wp skip on EL2
> > > 	- fix microbench test
> > > 	- add to CI
> > > 	- applied the fixup's Drew talked about in v5
> > > 
> > > About this failure in micro-bench:
> > > 
> > > 	- micro-bench hits the assert in gicv3_lpi_alloc_tables()
> > > 		lib/arm/gic-v3.c:183: assert failed: gicv3_data.redist_base[cpu]: Redistributor for cpu0 not initialized. Did cpu0 enable the GIC?	
> > > 
> > > in micro-bench.c lpi_prep() looks like this:
> > > 
> > > 	static bool lpi_prep(void)
> > > 	{
> > > 	        struct its_collection *col1;
> > > 	        struct its_device *dev2;
> > > 	        if (!gicv3_its_base())
> > > 	                return false;
> > > 
> > > QEMU by default runs with gic-v2, so gicv3_its_base() should have returned
> > > false and the lpi tests skipped. However because of the overflow mentioned in
> > > the first commit (new to v6), the its_data.base was being overwritten to a
> > > non-zero value.  This meant the test tried to do gic-v3 things on a gic-v2.
> > 
> > Nice work on the micro-bench/gic issue. Unfortunately the timer test still
> > times out, and that breaks the new CI since that test is listed.
> 
> Can you drop the timer from the CI commit while I investigate? It passes here,
> but I can try look more into. If everything else is working will be nice to
> merge.
>

Done. Hopefully you'll find a way to reproduce the timer issue and get it
fixed soon.

Thanks,
drew